modelbench日本語

Claude Sonnet 4.5

AnthropicLegacy

Input price
$3.00
Per 1 million tokens (USD)
Output price
$15.00
Per 1 million tokens (USD)
Effort levels
not supported
Benchmark results
9total
Benchmark types: 9

Model information

API model ID
claude-sonnet-4-5
release date
Unknown
effort
not supported
Effort source
Not supported: Anthropic effort documentation (platform.claude.com/docs/en/build-with-claude/effort) confirms effort is not supported (researched 2026-11-12).
Claude Code
available
Codex
unavailable
Availability notes
`sonnet` resolves to Sonnet 4.5 on Bedrock/Agent Platform/Foundry. effort not supported. Codex's official model list (learn.chatgpt.com/docs/models) contains only OpenAI models, so codex is false (connections via a custom provider are not counted as official options).
Availability source
https://code.claude.com/docs/en/model-config

Price (USD per 1 million tokens)

Prices do not vary by effort; token consumption does.

input$3.00
output$15.00
cache read$0.30
cache write$3.75

cacheWritePerMTok is for a 5-minute cache write. 1h write $6. 200K context.

Claude Sonnet 4.5 | $3 / MTok | $3.75 / MTok | $6 / MTok | $0.30 / MTok | $15 / MTok

Source: https://platform.claude.com/docs/en/about-claude/pricing

Benchmark results

BenchmarkeffortScoreSource cost (metric)measured/reported bysourceConditions and extraction method
GPQA Diamondnot statednot stated73.7%—Epoch AIEpoch AI, Capabilities & benchmarking (benchmark_data.zip, gpqa_diamond.csv) CC BY 4.0third-party · Retrieved 2026-11-12Run by Epoch AI (third-party measurement). Source label "Claude Sonnet 4.5 (no thinking)". No effort suffix in Model version. Data retrieved 2026-11-12. No effort because effort is not stated in the source.transcribed from table
Source excerptgpqa_diamond.csv: Model version=claude-sonnet-4-5-20250929, Best score (across scorers)=0.7373737373737373, stderr=0.031353050095330834, Started at=2025-09-29T18:04:57.033ZLocation gpqa_diamond.csv, data row 221 (file line 321), id=fsZmzw8jjyUgMQGwxodBwN, Model version=claude-sonnet-4-5-20250929
OTIS Mock AIME 2024-2025not statednot stated35.6%—Epoch AIEpoch AI, Capabilities & benchmarking (benchmark_data.zip, otis_mock_aime_2024_2025.csv) CC BY 4.0third-party · Retrieved 2026-11-12Run by Epoch AI (third-party measurement). Source label "Claude Sonnet 4.5 (no thinking)". No effort suffix in Model version. Data retrieved 2026-11-12. No effort because effort is not stated in the source.transcribed from table
Source excerptotis_mock_aime_2024_2025.csv: Model version=claude-sonnet-4-5-20250929, Best score (across scorers)=0.35555555555555557, stderr=0.0721639236343101, Started at=2025-09-29T18:39:45.629ZLocation otis_mock_aime_2024_2025.csv, data row 230 (file line 340), id=dEVgttSw8u6Tt2RsPWTc3C, Model version=claude-sonnet-4-5-20250929
SimpleQA Verifiednot statednot stated23.7%—Epoch AIEpoch AI, Capabilities & benchmarking (benchmark_data.zip, simpleqa_verified.csv) CC BY 4.0third-party · Retrieved 2026-11-12Run by Epoch AI (third-party measurement). Source label "Claude Sonnet 4.5 (no thinking)". No effort suffix in Model version. Data retrieved 2026-11-12. No effort because effort is not stated in the source.transcribed from table
Source excerptsimpleqa_verified.csv: Model version=claude-sonnet-4-5-20250929, Best score (across scorers)=0.237, stderr=0.013454070462577935, Started at=2026-08-10T22:42:10.000ZLocation simpleqa_verified.csv, data row 98 (file line 148), id=MhqeuRVx4JcRfxPQDYPq7Z, Model version=claude-sonnet-4-5-20250929
Chess Puzzles(Epoch AI)not statednot stated6.0%—Epoch AIEpoch AI, Capabilities & benchmarking (benchmark_data.zip, chess_puzzles.csv) CC BY 4.0third-party · Retrieved 2026-11-12Run by Epoch AI (third-party measurement). Source label "Claude Sonnet 4.5 (no thinking)". No effort suffix in Model version. Data retrieved 2026-11-12. No effort because effort is not stated in the source.transcribed from table
Source excerptchess_puzzles.csv: Model version=claude-sonnet-4-5-20250929, Best score (across scorers)=0.06, stderr=0.023868325657594204, Started at=2026-10-13T13:43:54.000ZLocation chess_puzzles.csv, data row 96 (file line 147), id=VajBDnUCyX7uqAfMTbi63j, Model version=claude-sonnet-4-5-20250929
Epoch Capabilities Index(ECI)not statednot stated147.63—Epoch AIEpoch AI, epoch_capabilities_index/eci_scores.csv (CC BY 4.0)third-party · Retrieved 2026-11-12A per-model metric that combines results across multiple benchmarks and multiple efforts (not per effort). Calculated by Epoch AI.transcribed from table
Source excerpteci_scores.csv: Model=Claude Sonnet 4.5, eci=147.63, eci_ci_low=146.17, eci_ci_high=149.15Location epoch_capabilities_index/eci_scores.csv, data row 80 (file line 81), Model=Claude Sonnet 4.5
τ²-bench airline (v1.0.1)not statednot stated72.0%$0.30Average cost in USD to run one trajectory in this domainSierraτ²-bench leaderboard submission claude-sonnet-4-5_sierra_2026-02-26 (sierra-research/tau2-bench, MIT)third-party · Retrieved 2026-11-13τ²-bench leaderboard submission claude-sonnet-4-5_sierra_2026-02-26 (submission_type standard, tau2_bench_version 1.0.1). Submitter "Sierra", model_organization "Anthropic": Run by Sierra, which operates τ²-bench (third-party measurement). Verified on the leaderboard (trajectories available, prompts not modified, no questions omitted). The value is pass^1 (%). The source label for effort, "enabled", is not in the official effort list, so no effort (reasoning_effort "enabled"). User simulator gpt-5.2 (reasoning_effort low). Evaluation date 2026-02-24. Notes summary: 4 trial runs; seed 300; extended thinking enabled.transcribed from table
Source excerptairline.pass_1=72.0; airline.cost=0.29646Location In submissions/claude-sonnet-4-5_sierra_2026-02-26/submission.json: results.airline.pass_1 (line 22 of file), results.airline.cost (line 26). model_name "Claude Sonnet 4.5". methodology.notes (line 56)
τ²-bench banking_knowledge (v1.0.1)not statednot stated25.3%$4.05Average cost in USD to run one trajectory in this domainSierraτ²-bench leaderboard submission claude-sonnet-4-5_sierra_2026-02-26 (sierra-research/tau2-bench, MIT)third-party · Retrieved 2026-11-13τ²-bench leaderboard submission claude-sonnet-4-5_sierra_2026-02-26 (submission_type standard, tau2_bench_version 1.0.1). Submitter "Sierra", model_organization "Anthropic": Run by Sierra, which operates τ²-bench (third-party measurement). Verified on the leaderboard (trajectories available, prompts not modified, no questions omitted). The value is pass^1 (%). The source label for effort, "enabled", is not in the official effort list, so no effort (reasoning_effort "enabled"). User simulator gpt-5.2 (reasoning_effort low). Search configuration retrieval_config "terminal". Evaluation date 2026-02-24. Notes summary: rescored on 2026-07-15 following a scoring correction in v1.0.1; 4 trial runs; seed 300; extended thinking enabled.transcribed from table
Source excerptbanking_knowledge.pass_1=25.26; banking_knowledge.cost=4.04814Location In submissions/claude-sonnet-4-5_sierra_2026-02-26/submission.json: results.banking_knowledge.pass_1 (line 43 of file), results.banking_knowledge.cost (line 47). model_name "Claude Sonnet 4.5". methodology.notes (line 56)
τ²-bench retail (v1.0.1)not statednot stated72.4%$0.25Average cost in USD to run one trajectory in this domainSierraτ²-bench leaderboard submission claude-sonnet-4-5_sierra_2026-02-26 (sierra-research/tau2-bench, MIT)third-party · Retrieved 2026-11-13τ²-bench leaderboard submission claude-sonnet-4-5_sierra_2026-02-26 (submission_type standard, tau2_bench_version 1.0.1). Submitter "Sierra", model_organization "Anthropic": Run by Sierra, which operates τ²-bench (third-party measurement). Verified on the leaderboard (trajectories available, prompts not modified, no questions omitted). The value is pass^1 (%). The source label for effort, "enabled", is not in the official effort list, so no effort (reasoning_effort "enabled"). User simulator gpt-5.2 (reasoning_effort low). Evaluation date 2026-02-24. Notes summary: 4 trial runs; seed 300; extended thinking enabled.transcribed from table
Source excerptretail.pass_1=72.37; retail.cost=0.25494Location In submissions/claude-sonnet-4-5_sierra_2026-02-26/submission.json: results.retail.pass_1 (line 29 of file), results.retail.cost (line 33). model_name "Claude Sonnet 4.5". methodology.notes (line 56)
τ²-bench telecom (v1.0.1)not statednot stated84.9%$0.49Average cost in USD to run one trajectory in this domainSierraτ²-bench leaderboard submission claude-sonnet-4-5_sierra_2026-02-26 (sierra-research/tau2-bench, MIT)third-party · Retrieved 2026-11-13τ²-bench leaderboard submission claude-sonnet-4-5_sierra_2026-02-26 (submission_type standard, tau2_bench_version 1.0.1). Submitter "Sierra", model_organization "Anthropic": Run by Sierra, which operates τ²-bench (third-party measurement). Verified on the leaderboard (trajectories available, prompts not modified, no questions omitted). The value is pass^1 (%). The source label for effort, "enabled", is not in the official effort list, so no effort (reasoning_effort "enabled"). User simulator gpt-5.2 (reasoning_effort low). Evaluation date 2026-02-24. Notes summary: 4 trial runs; seed 300; extended thinking enabled.transcribed from table
Source excerpttelecom.pass_1=84.87; telecom.cost=0.48769Location In submissions/claude-sonnet-4-5_sierra_2026-02-26/submission.json: results.telecom.pass_1 (line 36 of file), results.telecom.cost (line 40). model_name "Claude Sonnet 4.5". methodology.notes (line 56)

Related rankings

References